Researchers at Stanford Medicine have developed a new customizable AI tool called PathChat, which can be trained by pathologists to identify different cells and offer diagnostic insights. This 'human-in-the-loop' framework aims to overcome the limitations of off-the-shelf AI models in pathology. The AI assistant, featured in articles by Nature, Stat News, and VentureBeat, has shown to be better and faster than both human pathologists and standalone AI systems. PathChat can analyze microscopy images, converse with pathologists about tumors, and provide diagnoses. The tool's performance on diagnostic questions and responses in pathology has been highlighted, showcasing its potential to revolutionize the field. The research involved contributions from BrighamWomens and was described in natBME.